DNA repair

With time, organic polymers change their structure. When an RNA or a protein molecule is damaged, the cell simply degrades it and synthesises a new copy from DNA.

DNA has no such backup. It is the master copy, and there is nothing to make a fresh one from. So when DNA itself is damaged, the damage has to be repaired rather than replaced.

This note explains what counts as damage, how much of it happens, the mechanisms cells use to correct it, and the human diseases that follow when one of those mechanisms fails.

DNA damage is any chemical alteration of DNA which disrupts its normal structure.

There are five main types.

1. A single-stranded or double-stranded break in the phosphodiester backbone.
2. Loss of a base, creating a gap called an abasic or AP site (apurinic or apyrimidinic).
3. Chemical modification of a base, or the presence of uracil in DNA.
4. Inappropriate chemical interaction of a nucleotide, usually its base, with another nucleotide or another molecule.
5. Mismatch: a pair of otherwise normal but non-complementary nucleotides, or a different number of nucleotides in the two strands.

An AP site is created when a base is lost from a nucleotide. AP stands for apurine or apyrimidine, according to which kind of base has gone. The sugar and the phosphate remain in the backbone; only the base is missing.

Chemical modification of a base happens at particular positions within the purine and pyrimidine rings, and the common reactions are spontaneous hydrolysis, alkylation, oxidation, methylation and deamination.

DNA crosslinks are the fourth type. Covalent bonds form between bases that are opposite one another on the two strands, or between neighbouring bases on the same strand. Either way the crosslink disturbs DNA replication and transcription, because the two strands can no longer be separated normally.

DNA damage is far more widespread than we expect, even under optimal conditions. These are empirical frequencies for specific kinds of damage.

The conclusion is that the cell cannot rely on luck to keep its genome functional. Relying on luck is possible only for very small genomes, such as those of RNA viruses. The first cells that evolved on Earth were simple enough to cope that way, but no modern cell can.

To maintain their genomes, cells correct DNA damage by a process called DNA repair.

Damage and mutation are both chemical alterations of DNA and they are causally related, but they are not the same, and the distinction matters.

A mutation replaces the original structure with another structure which is also permitted for DNA. The mutated region "looks like the real thing": it may have lost its biological function, but it is physically and chemically perfect. Mutated DNA is therefore stable and replicates normally.

Damaged DNA, by contrast, does not replicate normally. The single exception is a mismatch, which is chemically normal on both sides.

The practical consequence is stark: mutations are often compatible with life, and damages almost never are.

In the best case DNA repair restores the original structure. When it is not possible to work out what that structure was, some chemically appropriate structure is supplied instead, even though it may be new. So mutations often result from the repair of damage. DNA repair should not be blamed for this. Damaged DNA must be repaired at any cost, even if the cost is a mutation.

Repair is carried out by enzymes helped by regulatory proteins. Some of those proteins monitor the DNA structure, and when they find damage they recruit the others to repair it.

The factors that cause DNA damage are called genotoxic factors, and because they cause mutations as well they are called mutagens.

The major mutagens are ionising radiation, UV light, temperature, and some chemicals that react with DNA. The chemical group includes reactive oxygen species, deaminating agents such as nitrous acid (HNO₂), and alkylating agents such as nitrosamines.

Mutations are divided into spontaneous ones, arising with no evident cause, and induced ones, caused by a mutagen. In truth the "spontaneous" mutations are also caused by mutagenic factors, but ubiquitous ones: UV light, oxygen, and the body temperature of 37 °C.

DNA is safe only in a deep freezer.

These are the abbreviations used in the scientific literature for the DNA damage repair pathways.

• Direct reversal
• MMR, mismatch repair
• BER, base excision repair
• NER, nucleotide excision repair
• HR, homologous recombination
• NHEJ, non-homologous end joining
• MMEJ, microhomology-mediated end joining

Broadly, the first four deal with damage confined to one strand, where the other strand survives as a template. The last three deal with double-strand breaks, where no intact template is left at the site.

Single-strand breaks are repaired easily. The cell is used to them, because they occur normally on the lagging strand during replication.

• If there is only a nick, DNA ligase repairs it.
• Larger single-strand gaps are first filled by a DNA polymerase and then sealed by a ligase.

The danger is what happens if one is left. If a replication fork meets an unrepaired single-strand break, a double-strand break is created and one arm of the replication fork is separated from the rest.

Some types of single-strand damage involving the bases can be repaired by chemically reversing the damage itself, with no excision and no resynthesis.

The best example is the dimer formed by two adjacent thymines under UV light. An enzyme called photolyase reverses the reaction, restoring the two original thymines.

Unfortunately, placental mammals have lost this enzyme. They can repair thymine dimers only by nucleotide excision repair.

Excision repair is used for a number of single-strand damages. The damaged area is removed, and the resulting gap is filled using the other DNA strand as a template.

It is the most widely used and most important type of repair, and it is subdivided into two categories:

• base excision repair;
• nucleotide excision repair.

Base excision repair corrects single-base lesions. These include most of the common damages: deamination, oxidation, alkylation, the presence of uracil in DNA, and AP sites, meaning nucleotides without a purine or pyrimidine base.

The first step is recognition of the defective base and its removal. This is done by enzymes called DNA glycosylases, so named because they cleave the N-glycosidic (glycosyl) bond between the base and the deoxyribose.

Note what this achieves. The defective base has been replaced with another type of damage, an AP site. When the damage was an AP site from the beginning, this first stage is of course omitted.

Then another enzyme, AP endonuclease, recognises the AP site and cuts the phosphodiester backbone on the 5' side of the hole, creating a nick.

From there the repair can follow two pathways, short-patch and long-patch. The short-patch pathway leads to a repair tract of a single nucleotide; the long-patch pathway produces a repair tract of at least two nucleotides.

Short-patch BER. Only the abasic nucleotide, the one that lost its base, is removed and replaced by a normal one. In eukaryotes this is done by DNA polymerase beta (pol β). Pol β is unable to cut off a normal nucleotide standing in front of it, because it has no 5' to 3' exonuclease activity, but it can remove an abasic nucleotide. The enzyme then connects the correct nucleotide. The damaged strand still carries a nick, but the bases are now perfect, and a DNA ligase anneals the nick.

Long-patch BER. The abasic nucleotide plus the next several normal nucleotides, between two and ten of them, are removed. In eukaryotes pol δ or pol ε displaces these nucleotides into a single-stranded flap, which is then cut off by flap endonuclease FEN1. These are the same enzymes that act in replication. The remaining nick is sealed by ligase.

It is not yet known why the cell uses short-patch BER in some cases and long-patch BER in others, for the same type of damage.

Nucleotide excision repair deals with single-strand damages larger than a single base, the standard example being a thymine dimer.

The sequence is:

1. Regulatory proteins recognise the damaged site.
2. Helicases join and unwind the DNA around it.
3. Endonucleases cleave the damaged strand on both the 5' and the 3' side of the damage, and the part containing the defective nucleotide or nucleotides is removed.
4. A DNA polymerase fills the resulting gap and a ligase seals the remaining nick, as in long-patch BER.

Four proteins carry out NER in E. coli: UvrA, UvrB, UvrC, and UvrD, which is a helicase.

The UvrA-UvrB complex scans for and identifies the site of the helix distortion. On identification UvrA dissociates, and UvrC binds UvrB, which is still at the site. The UvrB-UvrC complex cleaves around the damaged site, and UvrD unwinds this segment to remove it. DNA polymerase I and DNA ligase then fill the gap.

In humans, deficiencies in the NER proteins most notably result in xeroderma pigmentosum, characterised by a high sensitivity to UV light.

DNA damage is most dangerous during replication, and cells are most sensitive to mutagens at this time.

When the replication fork reaches an unrepaired damage it is stalled. An attempt is then made to continue: the replicase is replaced with a specialised translesion DNA polymerase, which is less demanding of its template and inserts some nucleotides opposite the damaged ones.

By acting this way the translesion polymerase can introduce point mutations. That risk is nevertheless preferable to recombinational repair, which is a dangerous last-resort mechanism.

If translesion synthesis fails, the replicated DNA molecule will carry a damage in its old strand and, opposite it, a gap in the new strand. The result is a double-strand break (DSB).

Double-strand breaks are a very serious damage. Left unrepaired, or repaired incorrectly, they may result in massive loss of genetic information, genomic rearrangements, or cell death. If there are two or more breaks, they can be joined in the wrong combination, leading to structural chromosomal mutations.

DSBs are caused by ionising radiation and some chemicals, for example reactive oxygen species.

Which mechanism the cell uses depends on whether a backup copy of the sequence is available, and therefore on the stage of the cell cycle.

• If the cell has another copy of the same DNA molecule, repair is by homologous recombination.
• If there is no backup copy, as in a haploid unreplicated genome, or it is not to hand, repair is by non-homologous end joining.

When translesion synthesis has failed and a double-strand break has formed, the only solution is so-called post-replication repair, which uses the sister DNA molecule, or the homologous chromosome, as a model.

The ends of the DNA at the double-strand break are processed by some extra cuts. This creates a kind of primer from which the broken strands can be extended. The undamaged strands of the sister or non-sister chromatid are then used as templates to restore the original sequence.

In fact, homologous recombinational repair is somatic crossing over.

HR is believed to be active only during the S and G2 phases of the cell cycle, which is when a homologous template is available as the sister chromatid.

Sometimes the rescue recombination does not use homologous sequences at all. In mammalian cells, double-strand breaks are in fact repaired predominantly by the non-homologous DNA end joining pathway, also called canonical NHEJ (C-NHEJ).

NHEJ modifies the broken DNA ends and ligates them together with little or no homology. The mechanism is: a little processing occurs around the site of the break, then the ends are stuck together with a polymerase and a ligase, while some other proteins hold the pieces together.

Because it needs no template, NHEJ has a flexibility that permits rescue recombination on a wide range of DNA-end configurations, and it is not restricted to a certain phase of the cell cycle.

The mechanism is hit-or-miss and very error-prone. But when the homologue is not available nearby there is no reference, so the cell cannot do much better than this.

This pathway for DSB repair is also known as alternative non-homologous end joining (Alt-NHEJ).

MMEJ uses microhomologous sequences, several base pairs long, to align the broken ends before joining them. The mechanism leads to deletions flanking the original break.

MMEJ is of great interest because of its potential to destabilise the genome through gene deletions and chromosomal rearrangements, particularly in cells deficient in the canonical repair pathways, including HR. The microhomologous sequences are defined as chromosomal breakpoints, which is why MMEJ events have a great impact on evolution and on the development of disease.

Trypanosoma brucei changes some of its genes by MMEJ in order to stay invisible to the human immune system.

Trypanosomes cycle between their insect vector and their mammalian hosts, where they colonise the blood, the fat and the skin, and eventually cross the blood-brain barrier in late-stage infection.

In the mammalian host, Trypanosoma cells are covered in a dense layer of a specific glycoprotein, the variant surface glycoprotein (VSG). This highly immunogenic VSG layer acts as a barrier, protecting the parasite from the host immune response.

The trypanosomes maintain a persistent infection by continuously escaping that response, switching to new variants of the VSG gene. The gene switching is performed by MMEJ. After each switch, the immune system needs additional time to react.

Single strand annealing is another mechanism for double-strand break repair. It uses homologous repeats to bridge the two DSB ends.

The repeats involved are usually longer than 25 base pairs. Because they flank the single DSB, annealing them together causes a deletion rearrangement between the repeats, so SSA is relatively mutagenic. It is one of the reasons for chromosomal rearrangements.

The distinction between SSA, MMEJ and NHEJ comes down to the different protein factors involved and to the extent of homology used:

A base mismatch occurs when a non-complementary nucleotide, or no nucleotide at all, stands opposite a nucleotide.

Mismatches arise during replication, for three reasons:

• the rare errors of the replicase;
• the ability of bases to form rare tautomeric forms, which pair by other rules;
• the not-so-rare errors of translesion synthesis.

Mismatches caused by a different number of nucleotides in the two strands have a different origin. They are caused by compounds with aromatic rings, for example the acridine dyes. These are called intercalating agents because they insert themselves between the DNA bases. During replication, an intercalating agent in the template strand can mislead the DNA polymerase into inserting an additional nucleotide in the new strand; an agent in the new strand can make the enzyme omit a nucleotide. Either way, frameshift mutations occur.

Unlike other damages, mismatches do not disturb replication. That is exactly the problem: replication with an unrepaired mismatch guarantees a point mutation in one of the two daughter DNA molecules.

The crucial part of mismatch repair is to find out which of the two variants is the mistaken one, that is, which is new. The old strand carries the original, correct variant. Once the new strand has been identified, it is made complementary to the old strand by mechanisms similar to excision repair.

The new strand is identified by two differences.

1. Uracil. Some uracil is always incorporated into the new DNA strand during replication, because deoxyuridine triphosphate is present in the cell as a precursor of deoxythymidine triphosphate.
2. Methylation. Normally, some nucleotides carry methyl groups. In the old strand the methyl groups are already present during replication, whereas in the new strand methyl groups are added a few minutes afterwards.

In other words, for a short period the new strands are undermethylated. This fact is used in a process called methyl-directed post-replicative mismatch repair.

Mismatch repair in bacteria uses a protein complex, MutS, MutH and MutL, to recognise a mismatch, for instance a G-T pair.

1. A loop of DNA forms between the methyl group on the parental strand and the mismatched base pair.
2. A nuclease removes only the newly synthesised DNA strand from the loop.
3. DNA polymerase re-synthesises the removed region, and ligase seals the strand.

A global response is the reaction of a cell to massive DNA damage induced by exposure to a mutagen.

In bacteria, large-scale DNA damage induces the so-called SOS response. It is based on about 20 repair proteins whose synthesis is induced by the damage itself. In the resting state the repressor LexA suppresses the SOS genes; when LexA is destroyed the SOS genes are activated. These enzymes repair DNA efficiently but less accurately, allowing more mutations than ordinary DNA repair does.

Eukaryotes have no SOS response. Their complex genomes would tolerate neither damage severe enough to justify an SOS response nor the mutation rates associated with it. The main response of a eukaryotic cell to DNA damage is instead to activate the cell cycle checkpoints: until the damage is repaired, cells in G1 are not allowed to initiate replication, and cells in G2 are not allowed to enter mitosis.

Biallelic mutations in the genes responsible for DNA repair, across NER, MMR, HR and NHEJ, cause cancer predisposition syndromes of childhood. DNA repair disorders are associated with cancer predisposition and with other problems in children.

Constitutional mismatch repair deficiency (CMMRD) syndrome carries a high risk of developing one or more types of cancer in children and young adults, in particular brain cancer, leukaemia or lymphoma, and colorectal cancer.

Ataxia-telangiectasia affects the nervous system, the immune system and other body systems. It is characterised by progressive difficulty in coordinating movements, the ataxia of the name, beginning in early childhood and usually before the age of 5.

Fanconi anaemia is an inherited bone marrow failure. Half of the patients are diagnosed before the age of 10. The disorder is associated with a progressive deficiency in the production of all blood cells, an increased risk of developing a cancer of the blood-forming cells, and tumours of the head, neck, skin, gastrointestinal system or genital tract. Mutations in more than 20 genes are related to Fanconi anaemia.

Bloom syndrome is an inherited disorder characterised by short stature, a skin rash that develops after exposure to the sun, and a greatly increased risk of cancer. Affected individuals can develop any type of cancer and often develop more than one type. They also have distinctive facial features.

The disease is provoked by a recessive mutation in a gene for a helicase. The result is too many errors in DNA, and too many attempts to rescue the chromosomes by sister chromatid exchange. That in turn leads to genomic instability. Metaphase cells from a patient with Bloom syndrome show frequent sister chromatid exchanges, which is how the condition is recognised down the microscope.

Patients with Werner syndrome show early general ageing: greying and thinning of the hair, atrophy of the skin, loss of cutaneous fat, and bilateral cataracts. The effect is dramatic enough that photographs of the same person at 15 and at 48, or at 8 and at 36, look like photographs of two different people separated by a lifetime.

The genetic problem is a mutation in the WRN gene, which is responsible for DNA repair by recombination. The WRN protein has several functional domains: an exonuclease domain, a helicase domain, RQC, HRDC, and a nuclear localisation signal. Over 70 insertions and deletions, missense and nonsense mutations, and splice mutations have been found across those domains.

NER is vital if cells are to endure the DNA-damaging action of UV radiation during everyday exposure to sunlight. Defects in the genes responsible for NER cause an autosomal recessive disorder known as xeroderma pigmentosum.

Even a minimal amount of sunlight causes severe skin lesions, and skin cancer later develops in multiple foci. Avoiding all exposure to the sun is the only way to prolong life, which is why the patients are sometimes called "children of the night".

• Damage is not a mutation. A mutation is a chemically perfect but altered structure that replicates normally; damage is a chemically wrong structure that does not. Repair of damage is often what produces the mutation.
• Damage is constant, not exceptional: 55,000 single-strand breaks and 8,000 depurinations per mammalian cell per day.
• Single-strand damage is repairable because the other strand is an intact template. That is what direct reversal, BER, NER and mismatch repair all rely on.
• BER handles single-base lesions and works through glycosylase, AP endonuclease, then short-patch (one nucleotide, pol β) or long-patch (2 to 10 nucleotides, pol δ or ε plus FEN1).
• NER handles bulky lesions such as thymine dimers, and its failure causes xeroderma pigmentosum.
• Double-strand breaks have no intact template at the site. HR is accurate but needs a sister chromatid, so it works only in S and G2; NHEJ works at any time but is error-prone, and it is the predominant pathway in mammalian cells.
• Mismatch repair exists to answer one question: which strand is new? The answer is the undermethylated one.
• Bacteria answer massive damage with the SOS response; eukaryotes answer it with cell cycle checkpoints.